This header was originally copied from the Linux kernel. However, these
days all fixed-width integers are defined in stdint.h, and all of the
other typedefs in this file are kernel-specific and aren't used
anywhere, so we can drop it.
Change-Id: I6ee7acb5e12f4b4b7c4325cedcfee36b93ab6a3d
Signed-off-by: Jacob Garber <jgarber1@ualberta.ca>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/37257
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
All targets now have the _cbmem_top_ptr symbol populated via calling
arguments or in the nvidia/tegra210 case worked around by populating
it with cbmem_top_chipset explicitly at the start of ramstage, so the
Kconfig guarding this behavior can be removed.
Change-Id: Ie7467629e58700e4d29f6e735840c22ed687f880
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36422
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Michael Niewöhner
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This solution is very generic and can in principle be implemented on
all arch/soc.
Instead trying to figure out which files can be removed from stages
and which cbmem_top implementations need with preprocessor, rename all
cbmem_top implementation to cbmem_top_romstage.
Mechanisms set in place to pass on information from rom- to ram-stage
will be placed in a followup commit.
Change-Id: If31f0f1de17ffc92c9397f32b26db25aff4b7cab
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36145
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
These platforms return to romstage from FSP only after
already having torn CAR down. A copy of the entire CAR
region is available and discoverable via HOB.
Previously, CBMEM console detected on-the-fly that CAR
migration had happened and relocated cbmem_console_p
accoringlin with car_sync_var(). However, if the CAR_GLOBAL
pointing to another object inside CAR is a relative offset
instead, we have a more generic solution that can be used
with timestamps code as well.
Change-Id: Ica877b47e68d56189e9d998b5630019d4328a419
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35140
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
As discussed on the mailing list and voted upon, the coreboot project
is going to move the majority of copyrights out of the headers and into
an AUTHORS file. This will happen a bit at a time, as we'll be unifying
license headers at the same time.
Additional changes in this patch:
- Make sure files say that they're part of the coreboot project
- Move descriptions below the license header
- Fix some spacing
Signed-off-by: Martin Roth <martin@coreboot.org>
Change-Id: Id5f26a73f0df366651c076a94975bce1fb321b70
Reviewed-on: https://review.coreboot.org/c/coreboot/+/34605
Reviewed-by: Werner Zeh <werner.zeh@siemens.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
There are only minimal differences between the architecture specific
stdint.h implementations, so let's tidy them up and merge them together
into a single file. In particular,
- Use 'unsigned long' for uintptr_t. This was already the case for x86
and riscv, while arm and mips used 'unsigned int', and arm64 and ppc64
used 'unsigned long long'. This change allows using a single integer
type for uintptr_t across all architectures, and brings it into
consistency with the rest of the code base, which generally uses
'unsigned long' for memory addresses anyway. This change required
fixing several assumptions about integer types in the arm code.
- Use _Bool as the boolean type. This is a specialized boolean type that
was introduced in C99, and is preferrable over hacking booleans
using integers. romcc sadly does not support _Bool, so for that we
stick with the old uint8_t.
- Drop the least and fast integer types. They aren't used
anywhere in the code base and are an unnecessary maintenance burden.
Using the standard fixed width types is essentially always better anyway.
- Drop the UINT64_C() macro. It also isn't used anywhere and doesn't
provide anything that a (uint64_t) cast doesn't.
- Implement the rest of the MIN and MAX numerical limits.
- Use static assertions to check that the integer widths are correct.
Change-Id: I6b52f37793151041b7bdee9ec3708bfad69617b2
Signed-off-by: Jacob Garber <jgarber1@ualberta.ca>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/34075
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: HAOUAS Elyes <ehaouas@noos.fr>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
This patch cleans up remaining uses of raw boolean Kconfig values I
could find by wrapping them with CONFIG(). The remaining naked config
value warnings in the code should all be false positives now (although
the process was semi-manual and involved some eyeballing so I may have
missed a few).
Change-Id: Ifa0573a535addc3354a74e944c0920befb0666be
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31813
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
When <symbols.h> was first introduced, it only declared a handful of
regions and we didn't expect that too many architectures and platforms
would need to add their own later. However, our amount of platforms has
greatly expanded since, and with them the need for more special memory
regions. The amount of code duplication is starting to get unsightly,
and platforms keep defining their own <soc/symbols.h> files that need
this as well.
This patch adds another macro to cut down the definition boilerplate.
Unfortunately, macros cannot define other macros when they're called, so
referring to region sizes as _name_size doesn't work anymore. This patch
replaces the scheme with REGION_SIZE(name).
Not touching the regions in the x86-specific <arch/symbols.h> yet since
they don't follow the standard _region/_eregion naming scheme. They can
be converted later if desired.
Change-Id: I44727d77d1de75882c72a94f29bd7e2c27741dd8
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/31539
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Definitions of these types are arch-agnostic. Shared device
subsystem files cannot include arch/pci_ops.h for ARM
and arch/io.h for x86.
Change-Id: I6a3deea676308e2dc703b5e06558b05235191044
Signed-off-by: Elyes HAOUAS <ehaouas@noos.fr>
Reviewed-on: https://review.coreboot.org/c/29947
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Bounce buffers used to be used in those cases where the payload
might overlap coreboot.
Bounce buffers are a problem for rampayloads as they need malloc.
They are also an artifact of our x86 past before we had relocatable
ramstage; only x86, out of the 5 architectures we support, needs them;
currently they only seem to matter on the following chipsets:
src/northbridge/amd/amdfam10/Kconfig
src/northbridge/amd/lx/Kconfig
src/northbridge/via/vx900/Kconfig
src/soc/intel/fsp_baytrail/Kconfig
src/soc/intel/fsp_broadwell_de/Kconfig
The first three are obsolete or at least could be changed
to avoid the need to have bounce buffers.
The last two should change to no longer need them.
In any event they can be fixed or pegged to a release which supports
them.
For these five chipsets we change CONFIG_RAMBASE from 0x100000 (the
value needed in 1999 for the 32-bit Linux kernel, the original ramstage)
to 0xe00000 (14 Mib) which will put the non-relocatable x86
ramstage out of the way of any reasonable payload until we can
get rid of it for good.
14 MiB was chosen after some discussion, but it does fit well:
o Fits in the 16 MiB cacheable range coreboot sets up by default
o Most small payloads are well under 14 MiB (even kernels!)
o Most large payloads get loaded at 16 MiB (especially kernels!)
With this change in place coreboot correctly still loads a bzImage payload.
Werner reports that the 0xe00000 setting works on his broadwell systems.
Change-Id: I602feb32f35e8af1d0dc4ea9f25464872c9b824c
Signed-off-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-on: https://review.coreboot.org/28647
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Its spreading copies got out of sync. And as it is not a standard header
but used in commonlib code, it belongs into commonlib. While we are at
it, always include it via GCC's `-include` switch.
Some Windows and BSD quirk handling went into the util copies. We always
guard from redefinitions now to prevent further issues.
Change-Id: I850414e6db1d799dce71ff2dc044e6a000ad2552
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/28927
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Add a __always_inline macro that wraps __attribute__((always_inline))
and replace current users with the macro, excluding files under
src/vendorcode.
Change-Id: Ic57e474c1d2ca7cc0405ac677869f78a28d3e529
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/28587
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@google.com>
Masked ROMs are the silent killers of boot speed on devices without
memory-mapped SPI flash. They often contain awfully slow SPI drivers
(presumably bit-banged) that take hundreds of milliseconds to load our
bootblock, and every extra kilobyte of bootblock size has a hugely
disproportionate impact on boot speed. The coreboot timestamps can never
show that component, but it impacts our users all the same.
This patch tries to alleviate that issue a bit by allowing us to
compress the bootblock with LZ4, which can cut its size down to nearly
half. Of course, masked ROMs usually don't come with decompression
algorithms built in, so we need to introduce a little decompression stub
that can decompress the rest of the bootblock. This is done by creating
a new "decompressor" stage which runs before the bootblock, but includes
the compressed bootblock code in its data section. It needs to be as
small as possible to get a real benefit from this approach, which means
no device drivers, no console output, no exception handling, etc.
Besides the decompression algorithm itself we only include the timer
driver so that we can measure the boot speed impact of decompression. On
ARM and ARM64 systems, we also need to give SoC code a chance to
initialize the MMU, since running decompression without MMU is
prohibitively slow on these architectures.
This feature is implemented for ARM and ARM64 architectures for now,
although most of it is architecture-independent and it should be
relatively simple to port to other platforms where a masked ROM loads
the bootblock into SRAM. It is also supposed to be a clean starting
point from which later optimizations can hopefully cut down the
decompression stub size (currently ~4K on RK3399) a bit more.
NOTE: Bootblock compression is not for everyone. Possible side effects
include trying to run LZ4 on CPUs that come out of reset extremely
underclocked or enabling this too early in SoC bring-up and getting
frustrated trying to find issues in an undebuggable environment. Ask
your SoC vendor if bootblock compression is right for you.
Change-Id: I0dc1cad9ae7508892e477739e743cd1afb5945e8
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/26340
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Introduce new bootmem tags to allow more fine grained control over buffer
allocation on various platforms. The new tags are:
BM_MEM_RAMSTAGE : Memory where any kind of boot firmware resides and that
should not be touched by bootmem (by example: stack,
TTB, program, ...).
BM_MEM_PAYLOAD : Memory where any kind of payload resides and that should
not be touched by bootmem.
Starting with this commit all bootmem methods will no longer see memory
that is used by coreboot as usable RAM.
Bootmem changes:
* Introduce a weak function to add platform specific memranges.
* Mark memory allocated by bootmem as BM_TAG_PAYLOAD.
* Assert on failures.
* Add _stack and _program as BM_MEM_RAMSTAGE.
ARMv7 and ARMv8 specific changes:
* Add _ttb and _postram_cbfs_cache as BM_MEM_RAMSTAGE.
ARMv7 specific changes:
* Add _ttb_subtables as BM_MEM_RAMSTAGE.
Change-Id: I0c983ce43616147c519a43edee3b61d54eadbb9a
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/25383
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
In builds without CONFIG_VBOOT_SEPARATE_VERSTAGE, verstage files are
linked directly into the bootblock or the romstage. However, they're
still compiled with a separate "libverstage" source file class, linked
into an intermediate library and then linked into the final destination
stage.
There is no obvious benefit to doing it this way and it's unclear why it
was chosen in the first place... there are, however, obvious
disadvantages: it can result in code that is used by both libverstage
and the host stage to occur twice in the output binary. It also means
that libverstage files have their separate compiler flags that are not
necessarily aligned with the host stage, which can lead to weird effects
like <rules.h> macros not being set the way you would expect. In fact,
VBOOT_STARTS_IN_ROMSTAGE configurations are currently broken on x86
because their libverstage code that gets compiled into the romstage sets
ENV_VERSTAGE, but CAR migration code expects all ENV_VERSTAGE code to
run pre-migration.
This patch resolves these problems by removing the separate library.
There is no more difference between the 'verstage' and 'libverstage'
classes, and the source files added to them are just treated the same
way a bootblock or romstage source files in configurations where the
verstage is linked into either of these respective stages (allowing for
the normal object code deduplication and causing those files to be
compiled with the same flags as the host stage's files).
Tested this whole series by booting a Kevin, an Elm (both with and
without SEPARATE_VERSTAGE) and a Falco in normal and recovery mode.
Change-Id: I6bb84a9bf1cd54f2e02ca1f665740a9c88d88df4
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/18302
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
This change adds armv7-r support for all stages.
armv7-r is an ARM processor based on the Cortex-R series.
Currently, there is support for armv7-a and armv7-m and
armv7-a files has been modfied to accommodate armv7-r by
adding ENV_ARMV7_A, ENV_ARMV7_R and ENV_ARMV7_M constants
to src/include/rules.h.
armv7-r exceptions support will added in a later time.
Change-Id: If94415d07fd6bd96c43d087374f609a2211f1885
Signed-off-by: Hakim Giydan <hgiydan@marvell.com>
Reviewed-on: https://review.coreboot.org/15335
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
BUG=chrome-os-partner:49249
TEST=None. Initial code not sure if it will even compile
BRANCH=none
Change-Id: Ib0fccfe2d103710c006cb3950c65b11b8d596912
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: 9be5f58bb89ec43d4eb264c94c3f745dcade35dd
Original-Change-Id: If50efb55d4974dfcab07d3ae6488c2413b505a1f
Original-Signed-off-by: Varadarajan Narayanan <varada@codeaurora.org>
Original-Reviewed-on: https://chromium-review.googlesource.com/333301
Original-Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/14657
Reviewed-by: Julius Werner <jwerner@chromium.org>
Tested-by: build bot (Jenkins)
